CN104778213B - A kind of social networks recommendation method based on random walk - Google Patents
A kind of social networks recommendation method based on random walk Download PDFInfo
- Publication number
- CN104778213B CN104778213B CN201510121263.2A CN201510121263A CN104778213B CN 104778213 B CN104778213 B CN 104778213B CN 201510121263 A CN201510121263 A CN 201510121263A CN 104778213 B CN104778213 B CN 104778213B
- Authority
- CN
- China
- Prior art keywords
- user
- article
- scoring
- migration
- social networks
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Abstract
The present invention relates to a kind of, and the social networks based on random walk recommends method, includes the following steps:1) a source user u and article i to be predicted is selected in social networks, sets the maximum value of migration step number k as 6;2) random walk is proceeded by by starting point of source user u in social networks, reaches user ukWhen, judge user ukWhether scoring is had to article i to be predicted;3) according to user ukAll articles that scored setThe probability for this time stopping migration being calculated with step number k4) a substitution goods j, record user u are selectedkScoring to substitution goods j is5) in social networks with ukThe user's set being connected directlyThe node users u of middle selection next step random walkk+1;6) according to the scoring recorded, scoring rs of the prediction source user u to article i to be predictedu,i, and article i to be predicted is recommended into source user u.Compared with prior art, the present invention has many advantages, such as that accuracy rate is high, coverage rate is wide, method is advanced.
Description
Technical field
The present invention relates to computer application technologies, recommend more particularly, to a kind of social networks based on random walk
Method.
Background technology
As problem of information overload is increasingly severe on network, how user rapidly and accurately finds the information of oneself needs
It is faced with prodigious challenge.The appearance of commending system alleviates this problem to a certain extent.Commending system is mainly analyzed
The historical behavior of user analyzes user preference and using the relationship between user, finally makes recommendation to user.It pushes away at present
Recommending recommendation method traditional in system has collaborative filtering recommending (Collaborative Filtering), content-based recommendation
(Content-based Recommendations) etc., it is fairly good that matrix decomposition is proved to effect in Netflix matches.So
And there is many problems for traditional recommendation method, are difficult to obtain to compare when the user less to some historical behavior, which does, to be recommended
Good effect, here it is the cold start-up problems in commending system.And between its real user it is also to pass there is trusting relationship
The commending system of system does not consider these factors.
In recent years, the recommendation based on social networks was a research hotspot.Recommendation based on social networks in a model can
Degree of belief between measure user, studying the verified recommendation from trusted user can more allow user to be received.As long as one
A user belongs to a social networks, there is the user being connected directly, then commending system can make recommendation, therefore is based on society
Hand over the recommendation of network that can significantly improve the coverage rate of commending system.Degree of belief in social networks between user can be divided into aobvious
Show degree of belief and implicit trust degree.Show that degree of belief refers to the degree of belief clearly indicated by user, and implicit trust degree is system
It is derived according to some user's history behaviors, such as a-c cycle, user's common friend number etc. between user.Trust simultaneously
Degree is divided into as direct degree of belief and indirect degree of belief, and direct degree of belief refers to the degree of belief between the user being connected directly, and
It is the degree of belief generated between the user not being connected directly at two by the propagation of direct degree of belief to connect degree of belief.Degree of belief
Propagation is a major issue during social networks is recommended.Massa proposes to propagate degree of belief using multiplication, and also has and consider
The maximum distance of belief propagation and minimum trust threshold values.Belief propagation distance is bigger, and degree of belief will decay.In recommendation
It can solve the problems, such as the cold start-up of commending system to a certain extent using degree of belief.Inay Ha propose a kind of combination user it
Between relationship and traditional Collaborative Filtering Recommendation Algorithm, calculate the weights between user first, according to six degree of separation theorems,
All paths between two users are found out in the figure being made of user to calculate weights between the two.Finally utilize use
Family collaborative filtering, which is done, considers this weights factor when recommending, the experimental results showed that this method in accuracy on obtain it is certain
Raising.J.Golbeck proposes TidalTrust models, when predicting scoring of the source user to article in this model, adopts
With the strategy of breadth traversal, find out with source user distance it is nearest and have a user that scoring records to the article, and by these
User, which does the scoring of the article with the result that the degree of belief between user and source user is multiplied, to be polymerize, finally as point of prediction
Number.As long as but the shortcomings that this model be with source user distance slightly remote user be all not accounted in addition they between
Degree of belief it is bigger.MoleTrust is for TidalTrust, the difference lies in that having scoring to certain article finding
It is used as limitation provided with a depth capacity when user of record.Mohsen Jamali et al. propose a Random Graph trip
Walk model, belief propagation controlled using Random Graph migration strategy, but when migration distance farther out when consider to use homologue
Product generate many errors when substituting target item;Hao Ma indicate that there is Some features in social networks:User believes
Appoint and the Interest Similarity positive incidence of user, the Interest Similarity of friend relation and user do not have positive incidence, Yong Huhe
Interest between its good friend is different;Nicola Barbieri et al. propose a random topic model (WTFW), WTFW
It can predict to whether there is side between user, and social or topic interest explanation can be made to the side of prediction.
However when found in social networks commending system with the relevant user of source user, if it is considered that user spacing
From longer, then the coverage rate for generating recommendation is higher, but noise is bigger simultaneously.This is always that the social networks based on degree of belief pushes away
Where problem in recommending.We propose a model, the trust between the model energy measure user using Random Graph migration strategy
Degree, and perform well in terms of coverage rate and error.
Invention content
High, covering that it is an object of the present invention to overcome the above-mentioned drawbacks of the prior art and provide a kind of accuracys rate
Rate is wide, the advanced social networks based on random walk of method recommends method.
The purpose of the present invention can be achieved through the following technical solutions:
A kind of social networks recommendation method based on random walk, includes the following steps:
1. a kind of social networks based on random walk recommends method, which is characterized in that include the following steps:
1) in social networks select a source user u and article i to be predicted, using source user u as starting point proceed by with
Machine migration, and the maximum value of migration step number k is set as 6;
2) as arrival user ukWhen, judge user ukWhether scoring is had to article i to be predicted, if so, recording the scoring and beingStop migration, and carry out step 6), if it is not, then carrying out step 3);
3) according to user ukAll articles that scored setThe probability for this time stopping migration being calculated with step number kAnd withProbability carry out step 4), withProbability carry out step 5);
4) in user ukAll articles that scored setOne substitution goods j of middle selection, record user ukTo generation
Scoring for article j isStop migration, and carries out step 6);
If 5) migration step number k is less than 6, k=k+1, continue this migration, and in social networks with ukIt is connected directly
User setThe node users u of middle selection next step random walkk+1, and return to step 2), if migration step number k etc.
In 6, then stop migration, carries out step 6);
6) according to the scoring recorded, scoring rs of the prediction source user u to article i to be predictedu,i, and by article to be predicted
I recommends source user u.
Probability in the step 3)Computational methods be:
31) it calculates in user ukAll articles that scored setIn each article m and article i to be predicted it
Between article similarity sim (m, i), the calculating formula of sim (m, i) is:
Wherein, UCm,iTo have the set of the user of scoring, u to article m and ik∈UCm,i,WithRespectively user
ukScoring to article m and i,For user ukTo the average mark of all articles that scored;
32) probability for stopping migration being calculated according to article similarity sim (m, i)Calculating formula be:
The condition of selection substitution goods j meets in the step 4):
Wherein, t (uk-1,uk) it is uk-1With ukMutual trust degree, V be this migration pass through all path node (u1,
u2...uk) set.
The node users u of selection next step in the step 5)k+1Probability be:
Wherein, w is and ukThe user's set being connected directlyIn user node.
Prediction source user u is to article i to be predicted scorings r in the step 6)u,iCalculating formula be:
Wherein, R*For (the u returned after multiple migrationk, j) and set,For user ukScoring to substitution goods j, ru,iIt is
User ukTo the scoring weighted sum that either other replacement users score to substitution goods j of article i.
Compared with prior art, the present invention has the following advantages:
One, accuracy rate is high:It is provided with rational threshold value when due to considering similar article, and only considers the use that source user is trusted
The scoring article at family, therefore the accuracy rate of recommendation can be effectively improved.
Two, coverage rate is wide:Belief propagation distance is controlled according to random walk, the covering of recommendation can be effectively improved
Rate.
Three, method is advanced:The neighborhood that can be extended one's service using belief propagation can solve system to a certain extent
Cold start-up problem.
Description of the drawings
Fig. 1 is flow chart of the method for the present invention.
Specific implementation mode
The present invention is described in detail with specific embodiment below in conjunction with the accompanying drawings.
Embodiment:
A kind of social networks recommendation method based on random walk, comments article i to be predicted prediction source user u
Point, multiple migration is carried out according to following strategy, is included the following steps:
1) graph structure of social networks is built according to the trusting relationship between user, vertex is user, between vertex
Relationship is the degree of belief between user, a source user u and article i to be predicted is selected in social networks, and with source user u
Random walk is proceeded by for starting point, and sets the maximum value of migration step number k as 6;
2) as migration kth step, k=1,2,3 ... 6, arrival user ukWhen, judge user ukWhether to article i to be predicted
There is scoring, if so, recording the scoring and beingStop migration, and carry out step 6), if it is not, then carrying out step 3);
3) according to user ukAll articles that scored setThe probability for this time stopping migration being calculated with step number kAnd withProbability carry out step 4), withProbability carry out step 5), probabilityCalculating side
Method is:
31) it calculates in user ukAll articles that scored setIn each article m and article i to be predicted it
Between article similarity sim (m, i), the calculating formula of sim (m, i) is:
Wherein, UCm,iTo have the set of the user of scoring, u to article m and ik∈UCm,n,WithRespectively user
ukScoring to article m and i,For user ukTo the average mark of all articles that scored;
32) probability for stopping migration being calculated according to article similarity sim (m, i)Calculating formula be:
4) in user ukAll articles that scored setOne substitution goods j of middle selection, record user ukTo generation
Scoring for article j isStop migration, and carry out step 6), the condition of substitution goods j is selected to meet:
Wherein, t (uk-1,uk) it is uk-1With ukMutual trust degree, V be this migration pass through all path node (u1,
u2...uk) set;
The result of this time migration is record user ukScoring to j is not only allowed for when considering substitution goods here
Similarity between article, and the degree of belief between reasonable consideration user, therefore can effectively improve accuracy rate.
If 5) migration step number k is less than 6, k=k+1, continue this migration, and in social networks with ukIt is connected directly
User setThe node users u of middle selection next step random walkk+1, return to step 2), if migration step number k is equal to 6,
Then stop migration, carries out step 6), select the node users u of next stepk+1Probability be:
Wherein, w is and ukThe user's set being connected directlyIn user node;
Since the degree of belief between user, spread scope can be utilized to be not limited solely to the immediate neighbor of user, therefore
Coverage rate can be improved.
6) according to the scoring recorded, scoring rs of the prediction source user u to article i to be predictedu,i, and by article to be predicted
I recommends source user u, and prediction source user u is to article i to be predicted scorings ru,iCalculating formula be:
Wherein, R*For (the u returned after multiple migrationk, j) and set,For user ukScoring to substitution goods j, ru,iIt is
User ukTo the scoring weighted sum that either other replacement users score to substitution goods j of article i.
Claims (3)
1. a kind of social networks based on random walk recommends method, which is characterized in that include the following steps:
1) a source user u and article i to be predicted is selected in social networks, and random trip is proceeded by by starting point of source user u
It walks, and sets the maximum value of migration step number k as 6;
2) as arrival user ukWhen, judge user ukWhether scoring is had to article i to be predicted, if so, recording the scoring and being
Stop migration, and carry out step 6), if it is not, then carrying out step 3);
3) according to user ukAll articles that scored setThe probability for this time stopping migration being calculated with step number k
And withProbability carry out step 4), withProbability carry out step 5), probabilityComputational methods be:
31) it calculates in user ukAll articles that scored setIn each article m and article i to be predicted between
The calculating formula of article similarity sim (m, i), sim (m, i) is:
Wherein, UCm,iTo have the set of the user of scoring, u to article m and ik∈UCm,i,WithRespectively user ukIt is right
The scoring of article m and i,For user ukTo the average mark of all articles that scored;
32) probability for stopping migration being calculated according to article similarity sim (m, i)Calculating formula be:
4) in user ukAll articles that scored setOne substitution goods j of middle selection, record user ukTo sub
The scoring of product j isStop migration, and carry out step 6), the condition of substitution goods j is selected to meet:
Wherein, t (uk-1,uk) it is uk-1With ukMutual trust degree, V be this migration pass through all path node (u1,u2...uk)
Set;
If 5) migration step number k is less than 6, k=k+1, continue this migration, and in social networks with ukThe use being connected directly
Gather at familyThe node users u of middle selection next step random walkk+1, and return to step 2), if migration step number k is equal to 6,
Then stop migration, carries out step 6);
6) according to the scoring recorded, scoring rs of the prediction source user u to article i to be predictedu,i, and article i to be predicted is pushed away
It recommends and gives source user u.
2. a kind of social networks based on random walk according to claim 1 recommends method, which is characterized in that described
The node users u of selection next step in step 5)k+1Probability be:
Wherein, w is and ukThe user's set being connected directlyIn user node.
3. a kind of social networks based on random walk according to claim 1 recommends method, which is characterized in that described
Prediction source user u is to article i to be predicted scorings r in step 6)u,iCalculating formula be:
Wherein, R*For (the u returned after multiple migrationk, j) and set,For user ukScoring to substitution goods j, ru,iIt is user
ukTo the scoring weighted sum that either other replacement users score to substitution goods j of article i.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510121263.2A CN104778213B (en) | 2015-03-19 | 2015-03-19 | A kind of social networks recommendation method based on random walk |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510121263.2A CN104778213B (en) | 2015-03-19 | 2015-03-19 | A kind of social networks recommendation method based on random walk |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104778213A CN104778213A (en) | 2015-07-15 |
CN104778213B true CN104778213B (en) | 2018-10-26 |
Family
ID=53619677
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510121263.2A Expired - Fee Related CN104778213B (en) | 2015-03-19 | 2015-03-19 | A kind of social networks recommendation method based on random walk |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104778213B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109802859A (en) * | 2019-01-15 | 2019-05-24 | 腾讯科技(深圳)有限公司 | Nodes recommendations method and server in a kind of network |
Families Citing this family (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106100973A (en) * | 2016-06-07 | 2016-11-09 | 中国石油大学(华东) | A kind of personalized rubbish mail filtering method based on node similarity and defecator |
CN106530097B (en) * | 2016-10-11 | 2018-09-18 | 中国人民武装警察部队工程大学 | A kind of oriented social networks key propagation node discovery method based on random walk mechanism |
CN106713035B (en) * | 2016-12-23 | 2019-12-27 | 西安电子科技大学 | Congestion link positioning method based on grouping test |
CN107273396A (en) * | 2017-03-06 | 2017-10-20 | 扬州大学 | A kind of social network information propagates the system of selection of detection node |
CN107403390B (en) * | 2017-08-02 | 2020-06-02 | 桂林电子科技大学 | Friend recommendation method integrating Bayesian reasoning and random walk on graph |
CN107889082B (en) * | 2017-11-01 | 2020-04-14 | 南京邮电大学 | D2D device discovery method utilizing social relationship among users |
CN108133426B (en) * | 2017-12-25 | 2022-02-25 | 北京理工大学 | Social network link recommendation method |
CN110457592A (en) * | 2019-07-11 | 2019-11-15 | 天津大学 | A kind of social networks recommended method based on figure entropy |
CN110659416B (en) * | 2019-09-09 | 2022-05-17 | 秒针信息技术有限公司 | Recommendation method and recommendation device for browsing resources and readable storage medium |
CN112330405B (en) * | 2020-11-10 | 2023-11-24 | 深圳大学 | Recommendation method, terminal and storage medium for item set explicit feedback |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102779128A (en) * | 2011-05-10 | 2012-11-14 | 北京磊友信息科技有限公司 | Method and device for offline running of HTML5 (hypertext markup language 5) application program in mobile terminal |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7958120B2 (en) * | 2005-05-10 | 2011-06-07 | Netseer, Inc. | Method and apparatus for distributed community finding |
-
2015
- 2015-03-19 CN CN201510121263.2A patent/CN104778213B/en not_active Expired - Fee Related
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102779128A (en) * | 2011-05-10 | 2012-11-14 | 北京磊友信息科技有限公司 | Method and device for offline running of HTML5 (hypertext markup language 5) application program in mobile terminal |
Non-Patent Citations (1)
Title |
---|
基于随机游走模型的个性化信息推荐;王丽莎;《中国优秀硕士学位论文全文数据库信息科技辑》;20120715(第07期);第1页-第37页 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109802859A (en) * | 2019-01-15 | 2019-05-24 | 腾讯科技(深圳)有限公司 | Nodes recommendations method and server in a kind of network |
CN109802859B (en) * | 2019-01-15 | 2020-07-03 | 腾讯科技(深圳)有限公司 | Node recommendation method and server in network graph |
Also Published As
Publication number | Publication date |
---|---|
CN104778213A (en) | 2015-07-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104778213B (en) | A kind of social networks recommendation method based on random walk | |
CN104966125B (en) | A kind of article scoring of social networks and recommend method | |
CN103514255B (en) | A kind of collaborative filtering recommending method based on project stratigraphic classification | |
CN103793476B (en) | Network community based collaborative filtering recommendation method | |
Wang et al. | Diversified and scalable service recommendation with accuracy guarantee | |
CN103227731A (en) | Local computation method of node importance in complicated network based on improved structural hole | |
CN104268648B (en) | Merge user's ranking system of a variety of interactive information of user and user's subject information | |
CN102135999A (en) | User credibility and item nearest neighbor combination Internet recommendation method | |
US9147009B2 (en) | Method of temporal bipartite projection | |
CN107301583A (en) | It is a kind of that method is recommended based on user preference and the cold start-up trusted | |
CN105574541A (en) | Compactness sorting based network community discovery method | |
CN105678590A (en) | topN recommendation method for social network based on cloud model | |
CN109726319A (en) | A kind of user force analysis method based on interactive relation | |
CN103500168A (en) | Method and system for discovering communities in overlapped complex networks according to topology potential | |
Stanescu et al. | A hybrid recommender system: User profiling from keywords and ratings | |
CN109857928A (en) | User preference prediction technique based on polynary credit evaluation | |
CN108363752A (en) | User's social effectiveness analysis method of scale forecast is propagated based on microblogging | |
Pan et al. | A new reliability evaluation method for networks with imperfect vertices using BDD | |
Liang et al. | Personalized recommender systems integrating social tags and item taxonomy | |
CN103399919A (en) | Trust enhanced service push method based on social relation network | |
Wu et al. | Future link prediction in the blogosphere for recommendation | |
CN112597699B (en) | Social network rumor source identification method integrated with objective weighting method | |
CN111245657A (en) | Two network topology type directed link prediction indexes based on neighbor | |
Ma et al. | Fork probability analysis of PoUW consensus mechanism | |
CN106095882A (en) | A kind of personalization cloud recommendation method under cloudy platform environment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
EXSB | Decision made by sipo to initiate substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20181026 Termination date: 20210319 |
|
CF01 | Termination of patent right due to non-payment of annual fee |